An App defines a computation that will be executed asynchronously by Parsl. Apps are Python functions marked with a decorator which designates that the function will run asynchronously and cause it to return a Future instead of the result.

Apps can be one of three types of functions, each with their own type of decorator

  • @python_app: Most Python functions

  • @bash_app: A Python function which returns a command line program to execute

  • @join_app: A function which launches one or more new Apps

The intricacies of Python and Bash apps are documented below. Join apps are documented in a later section (see Join Apps).

Python Apps

def hello_world(name: str) -> str:
    return f'Hello, {name}!'


Python Apps run Python functions. The code inside a function marked by @python_app is what will be executed either locally or on a remote system.

Most functions can run without modification. Limitations on the content of the functions and their inputs/outputs are described below.

Rules for Function Contents

Parsl apps have access to less information from the script that defined them than functions run via Python’s native multiprocessing libraries. The reason is that functions are executed on workers that lack access to the global variables in the script that defined them. Practically, this means

  1. Functions may need to re-import libraries. Place the import statements that define functions or classes inside the function. Type annotations should not use libraries defined in the function.

import numpy as np

# BAD: Assumes library has been imported
def linear_model(x: list[float] | np.ndarray, m: float, b: float):
    return np.multiply(x, m) + b

# GOOD: Function imports libraries on remote worker
def linear_model(x: list[float] | 'np.ndarray', m: float, b: float):
    import numpy as np
    return np.multiply(x, m) + b
  1. Global variables are inaccessible. Functions should not use variables defined outside the function. Likewise, do not assume that variables created inside the function are visible elsewhere.

# BAD: Uses global variables
global_var = {'a': 0}

def counter_func(string: str, character: str = 'a'):
    global_var[character] += string.count(character)  # `global_var` will not be accessible

def counter_func(string: str, character: str = 'a'):
    return {character: string.count(character)}

for ch, co in good_global('parsl', 'a').result().items():
    global_var[ch] += co
  1. Outputs are only available through return statements. Parsl does not support generator functions (i.e., those which use yield statements) and any changes to input arguments will not be communicated.

# BAD: Assumes changes to inputs will be communicated
def append_to_list(input_list: list, new_val):

# GOOD: Changes to inputs are returned
def append_to_list(input_list: list, new_val) -> list:
    return input_list

Functions from Modules

The above rules assume that the user is running the example code from a standalone script or Jupyter Notebook. Functions that are defined in an installed Python module do not need to abide by these guidelines, as they are sent to workers differently than functions defined locally within a script.

Directly convert a function from a library to a Python App by passing it as an argument to python_app:

from module import function
function_app = python_app(function)

function_app will act as Parsl App function of function.

It is also possible to create wrapped versions of functions, such as ones with pinned arguments. Parsl just requires first calling update_wrapped() with the wrapped function to include attributes from the original function (e.g., its name).

from functools import partial, update_wrapped
import numpy as np
my_max = partial(np.max, axis=0, keepdims=True)
my_max = update_wrapper(my_max, max)  # Copy over the names
my_max_app = python_app(my_max)

The above example is equivalent to creating a new function (as below)

def my_max_app(*args, **kwargs):
    import numpy as np
    return np.max(*args, keepdims=True, axis=0, **kwargs)

Inputs and Outputs

Python apps may be passed any Python type as an input and return any Python type, with a few exceptions. There are several classes of allowed types, each with different rules.

  • Python Objects: Any Python object that can be saved with pickle or dill can be used as an import or output. All primitive types (e.g., floats, strings) are valid as are many complex types (e.g., numpy arrays).

  • Files: Pass files as inputs as a File object. Parsl can transfer them to a remote system and update the File object with a new path. Access the new path with File.filepath attribute.

    def read_first_line(x: File):
        with open(x.filepath, 'r') as fp:
            return fp.readline()

    Files can also be outputs of a function, but only through the outputs kwargs (described below).

  • Parsl Futures. Functions can receive results from other Apps as Parsl Future objects. Parsl will establish a dependency on the App(s) which created the Future(s) and start executing as soon as the preceding ones complete.

    def capitalize(x: str):
        return x.upper()
    input_file = File('text.txt')
    first_line_future = read_first_line(input_file)
    capital_future = capitalize(first_line_future)

    See the section on Futures for more details.

Learn more about the types of data allowed in the data section.


Any changes to mutable input arguments will be ignored.

Special Keyword Arguments

Some keyword arguments to the Python function are treated differently by Parsl

  1. inputs: (list) This keyword argument defines a list of input Futures or files. Parsl will wait for the results of any listed Futures to be resolved before executing the app. The inputs argument is useful both for passing files as arguments and when one wishes to pass in an arbitrary number of futures at call time.

def map_app(x):
    return x * 2

def reduce_app(inputs = ()):
    return sum(inputs)

map_futures = [map_app(x) for x in range(3)]
reduce_future = reduce_app(inputs=map_futures)

print(reduce_future.result())  # 0 + 1 * 2 + 2 * 2 = 6
  1. outputs: (list) This keyword argument defines a list of files that will be produced by the app. For each file thus listed, Parsl will create a future, track the file, and ensure that it is correctly created. The future can then be passed to other apps as an input argument.

def write_app(message, outputs=()):
    """Write a single message to every file in outputs"""
    for path in outputs:
        with open(path, 'w') as fp:
            print(message, file=fp)

to_write = [
    File(Path(tmpdir) / 'output-0.txt'),
    File(Path(tmpdir) / 'output-1.txt')
write_app('Hello!', outputs=to_write).result()
for path in to_write:
    with open(path) as fp:
        assert == 'Hello!\n'
  1. walltime: (int) This keyword argument places a limit on the app’s runtime in seconds. If the walltime is exceed, Parsl will raise an exception.


A Python app returns an AppFuture (see Futures) as a proxy for the results that will be returned by the app once it is executed. This future can be inspected to obtain task status; and it can be used to wait for the result, and when complete, present the output Python object(s) returned by the app. In case of an error or app failure, the future holds the exception raised by the app.

Options for Python Apps

The python_app() decorator has a few options which controls how Parsl executes all tasks run with that application. For example, you can ensure that Parsl caches the results of the function and executes tasks on specific sites.

@python_app(cache=True, executors=['gpu'])
def expensive_gpu_function():
    # ...

See the Parsl documentation for full details.


To summarize, any Python function can be made a Python App with a few restrictions

  1. Functions should act only on defined input arguments. That is, they should not use script-level or global variables.

  2. Functions must explicitly import any required modules if they are defined in script which starts Parsl.

  3. Parsl uses dill and pickle to serialize Python objects to/from apps. Therefore, Parsl require that all input and output objects can be serialized by dill or pickle. See Addressing SerializationError.

  4. STDOUT and STDERR produced by Python apps remotely are not captured.

Bash Apps

def echo(
    name: str,
    stdout=parsl.AUTO_LOGNAME  # Requests Parsl to return the stdout
    return f'echo "Hello, {name}!"'

future = echo('user')
future.result() # block until task has completed

with open(future.stdout, 'r') as f:

A Parsl Bash app executes an external application by making a command-line execution. Parsl will execute the string returned by the function as a command-line script on a remote worker.

Rules for Function Contents

Bash Apps follow the same rules as Python Apps. For example, imports may need to be inside functions and global variables will be inaccessible.

Inputs and Outputs

Bash Apps can use the same kinds of inputs as Python Apps, but only communicate results with Files.

The Bash Apps, unlike Python Apps, can also return the content printed to the Standard Output and Error.

Special Keywords Arguments

In addition to the inputs, outputs, and walltime keyword arguments described above, a Bash app can accept the following keywords:

  1. stdout: (string, tuple or parsl.AUTO_LOGNAME) The path to a file to which standard output should be redirected. If set to parsl.AUTO_LOGNAME, the log will be automatically named according to task id and saved under task_logs in the run directory. If set to a tuple (filename, mode), standard output will be redirected to the named file, opened with the specified mode as used by the Python open function.

  2. stderr: (string or parsl.AUTO_LOGNAME) Like stdout, but for the standard error stream.

  3. label: (string) If the app is invoked with stdout=parsl.AUTO_LOGNAME or stderr=parsl.AUTO_LOGNAME, this argument will be appended to the log name.


If the Bash app exits with Unix exit code 0, then the AppFuture will complete. If the Bash app exits with any other code, Parsl will treat this as a failure, and the AppFuture will instead contain an BashExitFailure exception. The Unix exit code can be accessed through the exitcode attribute of that BashExitFailure.

Execution Options

Bash Apps have the same execution options (e.g., pinning to specific sites) as the Python Apps.

MPI Apps

Applications which employ MPI to span multiple nodes are a special case of Bash apps, and require special modification of Parsl’s execution environment to function. Support for MPI applications is described in a later section.